Azasteroid Alkylators as Dual Inhibitors of AKT and ERK Signaling for the Treatment of Ovarian Carcinoma

(1) Background: Previous findings show that lactam steroidal alkylating esters display improved therapeutic efficacy with reduced toxicity. The aim of this study was to evaluate the anticancer activity of two newly synthesized aza-steroid alkylators (ENGA-L06E and ENGA-L08E) against human ovarian carcinoma cells, and consequently, the dual inhibition of RAS/PI3K/AKT and RAS/RAF/MEK/ERK signaling pathways, both of which are closely associated with ovarian cancer; (2) Methods: The in vitro cytostatic and cytotoxic effects of ENGA-L06E and ENGA-L08E were evaluated in a panel of five human ovarian cancer cell lines, as well as in in vivo studies. ENGA-L06E and ENGA-L08E, in addition to another two aniline-mustard alkylators, POPAM and melphalan (L-PAM), were utilized in order to determine the acute toxicity and antitumor efficacy on two human ovarian xenograft models. Also, in silico studies were performed in order to investigate the dual inhibition of ENGA-L06E and ENGA-L08E on RAS/PI3K/AKT and RAS/RAF/MEK/ERK signaling pathways; (3) Results: Both, in vitro and in vivo studies demonstrated that ENGA-L06E and ENGA-L08E were significantly more effective with a lower toxicity profile in comparison to POPAM and L-PAM alkylators. Moreover, in silico studies demonstrated that the two new aza-steroid alkylators could act as efficient inhibitors of the phosphorylation of AKT and ERK1/2 molecules; and (4) Conclusions: Both ENGA-L06E and ENGA-L08E demonstrated high anticancer activity through the inhibition of the PI3K-AKT and KRAS-ERK signaling pathways against human ovarian carcinoma, and thus constituting strong evidence towards further clinical development.


Introduction
Ovarian carcinoma (OC) is the most lethal gynecological malignancy due to drug resistance in current therapeutic approaches. There are several histological types of OC that are characterized by The analytical results of the inhibitory effects of ENGA-L06E and ENGA-L08E on ERK1/2 and AKT1/2 phosphorylation, in all studied cancer cell lines, and the corresponding kinetics according to the Michaelis Menten model are illustrated in Figures 1 and 2, as well as in Tables A1 and A2. According to the results, both compounds prevented the phosphorylation of ERK1/2 and AKT in all cell lines tested, effectively. The results on pERK1/2 and pAKT inhibitory effect induced by aza-steroidal alkylators ENGA-L06E and ENGA-L08E demonstrated similar patterns of kinetics (Vmax, Km). The highest effect on pERK1/2 inhibition resulted at 60 min and intervened in a dose and time-dependent relevance. In all the tested cell lines, taking together the kinetic parameters and IC50s (the concentrations of tested compounds that induce 50% inhibition of phosphorylation), the ENGA-L06E was shown to be more effective than ENGA-L08E in inhibiting ERK1/2 phosphorylation ( Figure 1, Table A1). However, both compounds provoked higher inhibition of pERK1/2 in UWB1.289 cells by about 85% at 60 min. On the other hand, the inhibition of AKT protein phosphorylation caused by ENGA-L08E was somewhat more potent than the inhibitory effect of ENGA-L06E, in all tested cell lines. The highest effect on pAKT inhibition resulted at 60 min and took place in a dose and time-dependent manner, similar to that of ERK1/2 phosphorylation ( Figure 2 and Table A2). Overall, ENGA-L08E exhibited a higher inhibitory effect on pAKT in OVCAR-3 cells while ENGA-L06E in SK-OV-3 and UWB1.289 cells (Figure 2). The inhibitory activity of both compounds on ERK1/2 and AKT phosphorylation was not affected Cancers 2020, 12, 1263 4 of 25 significantly by mutations on TP53, PIK3CA, microsatellite instability (MSI) status, as well as steroid receptor expression and sensitivity to cytotoxic drugs. However, KRAS mutation in OVCAR-5 cells may affect AKT phosphorylation, while MEK1 mutations, in OVCAR-3 cells, may derogate the inhibitory effect on ERK1/2 phosphorylation (Tables A1 and A2). Interestingly, both ENG-L06E and ENGA-L08E show an optimal inhibitory effect, on both ERK1/2 and AKT phosphorylation, in BRCA1-null UWB1.289 cells (Tables A1 and A2).
Cancers 2020, 12, x FOR PEER REVIEW 4 of 25 AKT phosphorylation was not affected significantly by mutations on TP53, PIK3CA, microsatellite instability (MSI) status, as well as steroid receptor expression and sensitivity to cytotoxic drugs. However, KRAS mutation in OVCAR-5 cells may affect AKT phosphorylation, while MEK1 mutations, in OVCAR-3 cells, may derogate the inhibitory effect on ERK1/2 phosphorylation (Tables  A1 and A2). Interestingly, both ENG-L06E and ENGA-L08E show an optimal inhibitory effect, on both ERK1/2 and AKT phosphorylation, in BRCA1-null UWB1.289 cells (Tables A1 and A2).   The inhibitory effect of ENGA-L06E and ENGA-L08E on the phosphorylation status of ERK1/2 and AKT, on both ovarian cancer cell lines, is presented in Figure 3. More specifically, the inhibition curves demonstrate t levels of percent expression of ERK1/2/GAPDH and AKT/GAPDH, which reflect more accurately the inhibition of pERK1/2 and pAKT in treated cells. In more detail, the phosphorylation of ERK1/2 and AKT was inhibited under the treatment of both compounds at 25 and 50 µM. As the inhibition curves depict, the two tested hybrid molecules yielded approximately 50-70% inhibition at 25 µM whilst phosphorylation of targeted proteins was almost totally inhibited (>85%) at higher concentrations (50 µM). Moreover, the inhibition of phosphorylation of ERK1/2 and AKT starts in 1 h while it gradually increases in 2 h after which it leads to a plateau ( Figure 3). Overall, ENGA-L08E induced a higher inhibition of AKT phosphorylation than ENGA-L06E, whereas ENGA-L06E demonstrated a higher inhibition on ERK1/2 phosphorylation than ENGA-L08E (in both cell lines ( Figure 3). Finally, preliminary experimental data (not shown) have shown that the tested azasteroid alkylators inhibit phosphorylation of AKT and ERK in the tumor xenograft tissue in vivo. More detailed quantitative studies on the inhibition of AKT and ERK phosphorylation and the effects on the expression of other predictive biomarkers on tumor cells, such as PD-L1, DNA mismatch repair (MMR) system, in correlation with the treatment schedules, time, and dose administration of the azasteroid alkylators in vivo, are currently under investigation. The inhibitory effect of ENGA-L06E and ENGA-L08E on the phosphorylation status of ERK1/2 and AKT, on both ovarian cancer cell lines, is presented in Figure 3. More specifically, the inhibition curves demonstrate t levels of percent expression of ERK1/2/GAPDH and AKT/GAPDH, which reflect more accurately the inhibition of pERK1/2 and pAKT in treated cells. In more detail, the phosphorylation of ERK1/2 and AKT was inhibited under the treatment of both compounds at 25 and 50 μΜ. As the inhibition curves depict, the two tested hybrid molecules yielded approximately 50-70% inhibition at 25 μΜ whilst phosphorylation of targeted proteins was almost totally inhibited (>85%) at higher concentrations (50 μΜ). Moreover, the inhibition of phosphorylation of ERK1/2 and AKT starts in 1 h while it gradually increases in 2 h after which it leads to a plateau ( Figure 3). Overall, ENGA-L08E induced a higher inhibition of AKT phosphorylation than ENGA-L06E, whereas ENGA-L06E demonstrated a higher inhibition on ERK1/2 phosphorylation than ENGA-L08E (in both cell lines ( Figure 3). Finally, preliminary experimental data (not shown) have shown that the tested azasteroid alkylators inhibit phosphorylation of AKT and ERK in the tumor xenograft tissue in vivo. More detailed quantitative studies on the inhibition of AKT and ERK phosphorylation and the effects on the expression of other predictive biomarkers on tumor cells, such as PD-L1, DNA mismatch repair (MMR) system, in correlation with the treatment schedules, time, and dose administration of the azasteroid alkylators in vivo, are currently under investigation.  ENGA-L06E, ENGA-L08E, L-PAM, and POPAM were evaluated for their in vivo acute toxicity, and such values are illustrated inTable 2. The conjugation of the alkylating agent with the aza-steroidal moiety resulted in significantly reduced acute toxicity effect and an increase of the LD 10 and LD 50 values. The LD 50 and LD 10 values induced by ENGA-L06E and ENGA-L08E were significantly decreased compared to POPAM and L-PAM. Moreover, ENGA-L08E had a significantly higher acute toxicity effect than ENGA-L06E (Table 2). In addition, ENGA-L06E and ENGA-L08E were administrated i.p. at the LD 10 dose and caused significant reduction in the size of SK-OV-3 ( Figure 4) and OVCAR-5 ( Figure 5) human xenograft tumor in mice. More specifically, 23 and 44 days after implantation, as well as 10 (d-10) and 16 days (d-16) after drug administration, the sizes of the primary tumors were 928 ± 16.4 mm 3 ( Figure 4) and 836 ± 67.4 mm 3 ( Figure 5) for the control group compared to 390 ± 21.4 mm 3 and 478 ± 18.3 mm 3 (Figure 4), as well as 113 ± 14 mm 3 and 220 ± 19.1 mm 3 ( Figure 5) for the groups receiving ENGA-L08E and ENGA-L06E, respectively.
The crystallographic structures of the proteins were used during the induced fit docking (IFD) simulations of hybrid steroid organic molecules ENGA-L06E and ENGA-L08E. Their crystal structures were prepared using the Protein Preparation Wizard [17], missing loops and side chains were prepared using Prime, ionized at pH 7.4 using PROPKA [18], and minimized using the OPLS3 force field [19]. The structures were designed using Maestro 10.2 and ionized at the target pH 7.4 using the Ligprep module [20]. Docking simulations were performed in order to elucidate the mechanism of action of the studied compounds at the molecular level, providing an interpretation of their biologic activity as anticancer drugs. The first step was to evaluate the active site for ERK1 and ERK2, a site that is targeted by the same co-crystallized ligand 38Z0, for both proteins ( Figure 6A). A grid was generated from the ERK1/2 co-crystallized inhibitor, and the 38Z0 was docked, resulting respectively, in satisfactory overlapping with the crystallographic data. Hydrogen atoms of ligands were omitted for clarity ( Figure 6B). ENGA-L06E was found to be less active in both ERK1 and ERK2 than the co-crystalized ligand 38Z0, whereas ENGA-L08E was more active than the ENGA-L06E. In addition, ENGA-L08E bound stronger than the co-crystalized ligand 38Z0 to ERK2 protein. It was also noticeable that in the ERK1 protein, the molecule ENGA-L08E bound efficiently compared to 38Z0 (Table A6).     Furthermore, treatment with ENGA-L08E and ENGA-L06E significantly increased the lifespan of SK-OV-3 as well as of OVCAR-5-bearing mice (Tables A5 and A6). On the other hand, POPAM and L-PAM showed no significant antitumor activity on SK-OV-3 (Table A3) and OVCAR-5-bearing mice (Table A4). Finally, the tumor response and anticancer activity assessment of ENGA-L06E, respectively, in satisfactory overlapping with the crystallographic data. Hydrogen atoms of ligands were omitted for clarity ( Figure 6B). ENGA-L06E was found to be less active in both ERK1 and ERK2 than the co-crystalized ligand 38Z0, whereas ENGA-L08E was more active than the ENGA-L06E. In addition, ENGA-L08E bound stronger than the co-crystalized ligand 38Z0 to ERK2 protein. It was also noticeable that in the ERK1 protein, the molecule ENGA-L08E bound efficiently compared to 38Z0 (Table A6). Stereochemistry is an important physicochemical property capable of explaining the differences in binding capacity. To this end, ENGA-L08E (with the hot pink color of C atoms) enters the cavity with the same direction as the 38Z0, in contrast to ENGA-L06E that approaches the binding site from the alkylating agent moiety while the steroidal moiety of lactam testosterone remains out of the cavity. Generally, both ENGA-L06E and ENGA-L08E (consisting of a modified steroidal skeleton) came from precursor steroids, testosterone and estrone, respectively, conjugated with the same alkylating agent. The former, due to its aromatic ring and the carboxyl group, which functions as a linker, increase the binding capacity of ENGA-L06E and ENGA-L08E via bonding to residues Lys71 with Halogen and pi-cation bond, and Asp184 with aromatic H-bond. Therefore, the binding efficiency of ENGA-L08E is influenced by the presence of estrone's aromatic ring in contrast to ENGA-L06E, in which the precursor steroid was testosterone ( Figure 7A,B). On the other hand, we evaluated the active site of AKT1 and AKT2 (in the grey grid), which are targeted by two different co-crystallized ligands, 0R4 and X39, respectively. In doing so, we overlapped the two proteins, and AKT1 (the dark green color) was found to be slightly bigger than the AKT2 (the yellow color), while there was a clear difference in the location of the active site where the co-crystallized ligands were located ( Figure 8A). In addition to the docking results for both proteins (Table A7), we decided to emphasize on AKT1 protein's cavity ( Figure 8B) given that both ENGA-L06E and ENGA-L08E seem to be inactive for AKT2 (Table A7). The specificity of the active center ( Figure 9A), in contrast to the active center of ERK1/2 protein, lies in the fact that it crosses the protein, and in order to inhibit it, the tested molecules should fit into the cavity the same way the co-crystallized ligand 0R4 does ( Figure  8B; Table A7). Stereochemistry is an important physicochemical property capable of explaining the differences in binding capacity. To this end, ENGA-L08E (with the hot pink color of C atoms) enters the cavity with the same direction as the 38Z0, in contrast to ENGA-L06E that approaches the binding site from the alkylating agent moiety while the steroidal moiety of lactam testosterone remains out of the cavity. Generally, both ENGA-L06E and ENGA-L08E (consisting of a modified steroidal skeleton) came from precursor steroids, testosterone and estrone, respectively, conjugated with the same alkylating agent. The former, due to its aromatic ring and the carboxyl group, which functions as a linker, increase the binding capacity of ENGA-L06E and ENGA-L08E via bonding to residues Lys71 with Halogen and pi-cation bond, and Asp184 with aromatic H-bond. Therefore, the binding efficiency of ENGA-L08E is influenced by the presence of estrone's aromatic ring in contrast to ENGA-L06E, in which the precursor steroid was testosterone ( Figure 7A,B). On the other hand, we evaluated the active site of AKT1 and AKT2 (in the grey grid), which are targeted by two different co-crystallized ligands, 0R4 and X39, respectively. In doing so, we overlapped the two proteins, and AKT1 (the dark green color) was found to be slightly bigger than the AKT2 (the yellow color), while there was a clear difference in the location of the active site where the co-crystallized ligands were located ( Figure 8A). In addition to the docking results for both proteins (Table A7), we decided to emphasize on AKT1 protein's cavity ( Figure 8B) given that both ENGA-L06E and ENGA-L08E seem to be inactive for AKT2 (Table A7). The specificity of the active center ( Figure 9A), in contrast to the active center of ERK1/2 protein, lies in the fact that it crosses the protein, and in order to inhibit it, the tested molecules should fit into the cavity the same way the co-crystallized ligand 0R4 does ( Figure 8B; Table A7).  After the evaluation of the active site for AKT1, we generated a specific grid for AKT1/2 in a similar way as described above. Subsequently, the hybrid molecules ENGA-L06E and ENGA-L08E were docked in the identified cavity and compared to the co-crystallized ligand OR4. The most active molecule appeared to be ENGA-L08E, whereas ENGA-L06E was also considered as comparable to the co-crystallized ligand (Table A7). A closer view of the docking solution revealed an extended hydrophobic region, as defined by residues Asn54, Gln59, Lys268 for ENGA-L08E and Gln203, Ser205, Thr211, Lys268 for ENGA-L06E ( Figure 9C). Although ENGA-L06E further appears to make more contacts with the residues in the active site of the protein, at the same time, it presents slightly higher energy than ENGA-L08E (Table A7) because of the lack of hydrogen bonds when compared to ENGA-L08E ( Figure 9C). The superimposition of the three molecules: 0R4 (in green color), ENGA-L06E (in purple color) and ENGA-L08E (in the hot pink color) revealed that the alkylating moiety of ENGA-L08E, and by extension the whole molecule, is precisely adjusted above the co-crystallized ligand in contrast to the ENGA-L06E molecule ( Figure 9B). More specifically, in Figure 10A and B, it is clearly represented how the ENGA-L08E molecule is adapted into the AKT1 protein's cavity. The  After the evaluation of the active site for AKT1, we generated a specific grid for AKT1/2 in a similar way as described above. Subsequently, the hybrid molecules ENGA-L06E and ENGA-L08E were docked in the identified cavity and compared to the co-crystallized ligand OR4. The most active molecule appeared to be ENGA-L08E, whereas ENGA-L06E was also considered as comparable to the co-crystallized ligand (Table A7). A closer view of the docking solution revealed an extended hydrophobic region, as defined by residues Asn54, Gln59, Lys268 for ENGA-L08E and Gln203, Ser205, Thr211, Lys268 for ENGA-L06E ( Figure 9C). Although ENGA-L06E further appears to make more contacts with the residues in the active site of the protein, at the same time, it presents slightly higher energy than ENGA-L08E (Table A7) because of the lack of hydrogen bonds when compared to ENGA-L08E ( Figure 9C). The superimposition of the three molecules: 0R4 (in green color), ENGA-L06E (in purple color) and ENGA-L08E (in the hot pink color) revealed that the alkylating moiety of ENGA-L08E, and by extension the whole molecule, is precisely adjusted above the co-crystallized ligand in contrast to the ENGA-L06E molecule ( Figure 9B). More specifically, in Figure 10A and B, it is clearly represented how the ENGA-L08E molecule is adapted into the AKT1 protein's cavity. The After the evaluation of the active site for AKT1, we generated a specific grid for AKT1/2 in a similar way as described above. Subsequently, the hybrid molecules ENGA-L06E and ENGA-L08E were docked in the identified cavity and compared to the co-crystallized ligand OR4. The most active molecule appeared to be ENGA-L08E, whereas ENGA-L06E was also considered as comparable to the co-crystallized ligand (Table A7). A closer view of the docking solution revealed an extended hydrophobic region, as defined by residues Asn54, Gln59, Lys268 for ENGA-L08E and Gln203, Ser205, Thr211, Lys268 for ENGA-L06E ( Figure 9C). Although ENGA-L06E further appears to make more contacts with the residues in the active site of the protein, at the same time, it presents slightly higher energy than ENGA-L08E (Table A7) because of the lack of hydrogen bonds when compared to ENGA-L08E ( Figure 9C). The superimposition of the three molecules: 0R4 (in green color), ENGA-L06E (in purple color) and ENGA-L08E (in the hot pink color) revealed that the alkylating moiety of ENGA-L08E, and by extension the whole molecule, is precisely adjusted above the co-crystallized ligand in contrast to the ENGA-L06E molecule ( Figure 9B). More specifically, in Figure 10A,B, it is clearly represented how the ENGA-L08E molecule is adapted into the AKT1 protein's cavity. The specificity of this particular active center ( Figure 9A), in contrast to the active center of ERK1/2 protein, lies in that it crosses the protein and in order to inhibit it, the tested molecules should fit into the cavity the same way the co-crystallized ligand 0R4 fits ( Figure 8B; Table A7). Subsequently, the hybrid molecules ENGA-L06E and ENGA-L08E were docked in the identified cavity and compared to the co-crystallized ligand. The most active molecule appeared to be ENGA-L08E, while ENGA-L06E was also considered comparable to the co-crystalized ligand (Table A7). A closer view of the docking solution in Figure 9C reveals an extended hydrophobic region, as defined by residues Asn54, Gln59, Lys268 for ENGA-L08E, and residues Gln203, Ser205, Thr211, Lys268 for ENGA-L06E. Although ENGA-L06E appears to make more contacts with the residues in the active site of the protein, at the same time, it presents slightly higher energy than ENGA-L08E (Table A7) probably because of the lack of hydrogen bonds ( Figure 9C). Complementary to the above data, the superimposition of the three molecules 0R4 (in green color), ENGA-L06E (in purple color) and ENGA-L08E (in the hot pink color) revealed that the alkylating moiety of ENGA-L08E, and by extension the whole molecule, is precisely adjusted above the co-crystalized ligand, in contrast to the ENGA-L06E molecule ( Figure 9B). More specifically, from the backside of the protein, it is clearly represented howtheENGA-L08E molecule is adapted into the AKT1 protein's cavity ( Figure 10B).
Cancers 2020, 12, x FOR PEER REVIEW 12 of 25 specificity of this particular active center ( Figure 9A), in contrast to the active center of ERK1/2 protein, lies in that it crosses the protein and in order to inhibit it, the tested molecules should fit into the cavity the same way the co-crystallized ligand 0R4 fits ( Figure 8B; Table A7). Subsequently, the hybrid molecules ENGA-L06E and ENGA-L08E were docked in the identified cavity and compared to the co-crystallized ligand. The most active molecule appeared to be ENGA-L08E, while ENGA-L06E was also considered comparable to the co-crystalized ligand (Table A7). A closer view of the docking solution in Figure 9C reveals an extended hydrophobic region, as defined by residues Asn54, Gln59, Lys268 for ENGA-L08E, and residues Gln203, Ser205, Thr211, Lys268 for ENGA-L06E.
Although ENGA-L06E appears to make more contacts with the residues in the active site of the protein, at the same time, it presents slightly higher energy than ENGA-L08E (Table A7) probably because of the lack of hydrogen bonds ( Figure 9C). Complementary to the above data, the superimposition of the three molecules 0R4 (in green color), ENGA-L06E (in purple color) and ENGA-L08E (in the hot pink color) revealed that the alkylating moiety of ENGA-L08E, and by extension the whole molecule, is precisely adjusted above the co-crystalized ligand, in contrast to the ENGA-L06E molecule ( Figure 9B). More specifically, from the backside of the protein, it is clearly represented howtheENGA-L08E molecule is adapted into the AKT1 protein's cavity ( Figure 10B).

Discussion
The new homo-aza-steroidal alkylating esters of POPAM were designed in the context of improving therapeutic efficacy against tumors resistant to alkylating drugs [7,9,21]. Lactam steroid alkylators (LSAs) combine two active compounds in a single molecule, incorporating modified steroids that bear the lactam (-NH-CO-) chemical moiety in one or more steroid rings, with the capacity to act as cytotoxic anticancer agents. These hybrid molecules were aimed to increase anticancer activity, reduce systemic toxicity, and improve the overall therapeutic ratio. Furthermore, LSAs induce a multi-targeted antitumor effect by inhibiting the function of critical cellular pathways important for cancer cell survival cancer cells [7][8][9][10][11][12][22][23][24][25].
Of particular importance, RAS/RAF/MEK/ERK and PI3K/PTEN/Akt/mTOR are two signaling pathways capable of regulating a wide array of cellular functions, including proliferation, differentiation, survival, metabolism, and migration. Various studies indicate that AKT and ERK are significantly implicated in cancer cell survival and that their synergism promotes cancer progression and metastasis, thus leading to poor therapeutic outcomes and tumor recurrence. Indeed, RAS/RAF/MEK/ERK and PI3K/PTEN/Akt/mTOR are important pathways. Their activation occurs in approximately 70% of all ovarian cancers. Furthermore, PI3K catalytic subunit alpha (PIK3CA) gene amplification has also been associated with genomic instability, p53 mutation, and a lack of response to chemotherapy. Evidence from ovarian cancer cell lines and animal models revealed that activation of the PI3K/AKT pathway might lead to chemotherapy resistance. Suppression of apoptosis contributed to platinum and taxane resistance. The chemotherapy-induced apoptosis has been restored by specific PI3K inhibitors in vitro and in vivo [26]. Consequently, targeting the regulation of downstream functions by AKT and ERK pathways is of great importance and therapeutic relevance in cancer [27,28]. Even though the downregulation of phosphorylated ERK or AKT may induce cell apoptosis or cell cycle arrest, the crosstalk between these two signaling pathways may counteract the effect of single inhibition. For instance, it has been shown that inhibition of the MEK/ERK pathway enhances AKT phosphorylation while ERK activation attenuates AKT phosphorylation. Therefore, inhibition of either pathway alone may not contribute significantly to the context of an anticancer therapy, whereas inhibition of both pathways may interrupt the transduction of survival signal(s), and consequently, achieve a significant therapeutic effect [29,30]. Previous research has demonstrated that the constitutive activation of AKT in various ovarian cancer cell lines (e.g., SK-OV-3, OVCAR-4, and OVCAR-5) promotes cell survival and resistance to cisplatin. Thus, Akt is a potential therapeutic target as its inhibition might increase cytotoxic sensitivity selectively in ovarian cancer that has a dependence on AKT kinase activity for survival [30]. On the other hand, therapies targeted towards MAPK/ERK components have various response rates in solid tumors, including ovarian cancer. The mechanisms responsible for the acquired chemoresistance in

Discussion
The new homo-aza-steroidal alkylating esters of POPAM were designed in the context of improving therapeutic efficacy against tumors resistant to alkylating drugs [7,9,21]. Lactam steroid alkylators (LSAs) combine two active compounds in a single molecule, incorporating modified steroids that bear the lactam (-NH-CO-) chemical moiety in one or more steroid rings, with the capacity to act as cytotoxic anticancer agents. These hybrid molecules were aimed to increase anticancer activity, reduce systemic toxicity, and improve the overall therapeutic ratio. Furthermore, LSAs induce a multi-targeted antitumor effect by inhibiting the function of critical cellular pathways important for cancer cell survival cancer cells [7][8][9][10][11][12][22][23][24][25].
Of particular importance, RAS/RAF/MEK/ERK and PI3K/PTEN/Akt/mTOR are two signaling pathways capable of regulating a wide array of cellular functions, including proliferation, differentiation, survival, metabolism, and migration. Various studies indicate that AKT and ERK are significantly implicated in cancer cell survival and that their synergism promotes cancer progression and metastasis, thus leading to poor therapeutic outcomes and tumor recurrence. Indeed, RAS/RAF/MEK/ERK and PI3K/PTEN/Akt/mTOR are important pathways. Their activation occurs in approximately 70% of all ovarian cancers. Furthermore, PI3K catalytic subunit alpha (PIK3CA) gene amplification has also been associated with genomic instability, p53 mutation, and a lack of response to chemotherapy. Evidence from ovarian cancer cell lines and animal models revealed that activation of the PI3K/AKT pathway might lead to chemotherapy resistance. Suppression of apoptosis contributed to platinum and taxane resistance. The chemotherapy-induced apoptosis has been restored by specific PI3K inhibitors in vitro and in vivo [26]. Consequently, targeting the regulation of downstream functions by AKT and ERK pathways is of great importance and therapeutic relevance in cancer [27,28]. Even though the downregulation of phosphorylated ERK or AKT may induce cell apoptosis or cell cycle arrest, the crosstalk between these two signaling pathways may counteract the effect of single inhibition. For instance, it has been shown that inhibition of the MEK/ERK pathway enhances AKT phosphorylation while ERK activation attenuates AKT phosphorylation. Therefore, inhibition of either pathway alone may not contribute significantly to the context of an anticancer therapy, whereas inhibition of both pathways may interrupt the transduction of survival signal(s), and consequently, achieve a significant therapeutic effect [29,30]. Previous research has demonstrated that the constitutive activation of AKT in various ovarian cancer cell lines (e.g., SK-OV-3, OVCAR-4, and OVCAR-5) promotes cell survival and resistance to cisplatin. Thus, Akt is a potential therapeutic target as its inhibition might increase cytotoxic sensitivity selectively in ovarian cancer that has a dependence on AKT kinase activity for survival [30]. On the other hand, therapies targeted towards MAPK/ERK components have various response rates in solid tumors, including ovarian cancer. The mechanisms responsible for the acquired chemoresistance in ovarian tumors are many, including efflux pumps, repair mechanisms, signaling survival pathways (e.g., PI3K/AKT, MAPK, estrogen signaling), etc. Specifically, PARP1 inhibition causes a loss of ERK2 stimulation by decreasing the activity of critical pro-angiogenic factors, including the vascular endothelial growth factor (VEGF) and the hypoxia-inducible factor (HIF) [17]. Thus, many natural and synthetic agents capable of suppressing the MAPK/ERK signaling pathway have been tested in order to overcome chemoresistance, and thus a series of targeted inhibitors have emerged [6,31]. Previous studies have shown that LSAs induce an anticancer response through the combination of the direct induction of cellular DNA damage with the inhibition of PARP enzyme(s), and consequently, the inhibition of DNA repair [12,25]. To these ends, inhibitors of nuclear PARP enzymes (e.g., PARP-1) have significantly improved clinical outcomes in ovarian cancer, especially in patients with BRCA1/2 gene mutations or additional homologous recombination DNA repair pathway deficiencies. Cancers with deficiency of DNA repair (by homologous recombination) are showing greater sensitivity to PARP inhibitors. Beyond BRCA1/2 gene mutations, the loss, perturbation, or overloading of function of various proteins involved in homologous recombination [e.g., ATM, CHEK2, BARD1, BRIP1, MRE11, RAD50, NBS1, RAD51C, RAD51D, RAD52, PALB2, and DNA-dependent protein kinase (DNA-PK)] have been suggested to be synthetically lethal with PARP inhibition. In that case, synthetic lethality could be potentially optimized when PARP inhibitors are combined with DNA-damaging agents, or with molecularly targeted agents capable of impairing DNA repair, such as the case of azasteroid alkylators [32,33].
The response to PARP inhibitors and platinum-based chemotherapy can be predicted by ADP-ribosylation (ADPRylation), which is also catalyzed by PARP enzymes, independently of the BRCA1/2 or homologous recombination deficiency (HRD) status [34]. In general, PARP-1 is activated by direct interaction with pERK2, which then amplifies ERK-signals, resulting in enhanced pERK2-catalyzed phosphorylation of target transcription factors and increased gene expression. Consequently, the pharmacologic inhibition of PARP can cause loss of ERK2 stimulation, thus blocking the ERK signaling network. Moreover, selective ERK inhibition induces tumor regression in MEK inhibitor-resistant models, and PARP inhibition could be similarly effective in MEK or ERK-resistant tumors. In fact, MEK, ERK, and PARP inhibitors, as well as their combinations, have recently entered in early phase clinical trials [35,36]. On the other hand, given that RAS mutant tumors are usually resistant to PARP inhibitors and other anticancer drugs, the inhibition of MEK and ERK can reverse PARP inhibitor resistance in KRAS mutant tumors, including ovarian cancer. Furthermore, the effects of treatment with combinations of PARP and MEK inhibitors are independent of BRCA1/2 and p53 mutation status, suggesting a generalized synergistic activity [37].
In general, the PI3K/AKT signaling pathway plays a critical role in the therapeutic management of ovarian cancer. More specifically, inhibition of PARP1 activity can lead to the upregulation of the PI3K/AKT pathway, thus promoting resistance to PARP inhibitors (PARPi). For instance, pharmacologic inhibition of PARP-1 promotes PI3K/AKT activity and mTOR signaling resulting in decreased cell death as well as PARPi response. Activation of the AKT survival pathway may counteract the cytotoxic effects of PARP inhibition, suggesting that the AKT pathway inhibition may enhance PARP inhibition in the context of antitumor therapy. As a result, combination(s) of small molecules targeting PARP, as well as the PI3K/AKT pathway, would potentially provide a greater clinical benefit [36,38]. On the other hand, the Ras-ERK and PI3K-mTOR signaling pathways intersect in order to regulate each other and to also co-regulate downstream functions. Both pathways can negatively regulate each other's activity, and cross-inhibition is often emerging. Moreover, when one pathway is chemically blocked, the other pathway can effectively be activated. Also, the Ras-ERK pathway cross-activates PI3K-AKT-mTORC1 signaling as long as Ras-GTP can directly bind and allosterically activate Phosphoinositide 3-kinase (PI3K). Finally, both pathways are often activated by genetic alterations in the upstream signaling molecules, which could further result in increased transcription of various genes promoting cell growth and malignant transformation. So, the rationale of dual inhibition of both of these pathways in suppressing ovarian cancer appears to be promising [39]. In fact, the extent and importance of this crosstalk and its significance in cancer therapeutics is becoming evident since AKT and ERK signaling are both aberrantly activated in a wide range of other human cancers [40]. Nevertheless, the utilization of dual-targeted inhibitors for PI3K/AKT/mTOR and RAS/ERK signaling pathways is a potentially effective therapeutic approach against ovarian cancer [41].
On the other hand, steroid hormones promote epithelial ovarian cancer growth, and their receptor expression is associated with disease outcome. Estrogen (ER-α/-β), and progesterone receptors PR(-A/-B) are frequently expressed in ovarian cancer with an important variability relating to histological subtype, grade, and stage. Hormone therapy is frequently used, but the significance of activity by specific agents or tumor characteristics is unknown. Although the activity of endocrine therapy in advanced ovarian cancer is considerable, limited clinical trials have shown modest response rates to endocrine therapy. However, a small subset of patients that respond very well was identified with few side effects [42][43][44]. The steroid components of ENGA-L06E and ENGA-L08E are testosterone and estrone derivatives, respectively. It has been reported that the steroidal moiety of homo-aza-steroid alkylators is binding to estrogen receptors (ER-α), in breast cancer cells with positive ER expression, and consequently was shown to be more sensitive to their cytostatic and cytotoxic effects [25,45]. However, both ENGA-L06E and ENGA-L08E compounds were shown significant anticancer activity against all human ovarian carcinoma cell lines tested and regardless of the steroid receptor expression.

Synthesis of Chemical Compounds
All modified hybrid steroidal esters and comparative alkylators were synthesized with methods that have been previously described [23]. More specifically, we have tested the two previously synthesized hybrid homo-azasteroid alkylators (e.g., ENGA-L06E and ENGA-L08E) along with the proper reference alkylator aniline mustards (e.g., POPAM and L-PAM). The lactam-steroidal carrier of ENGA-L06E is a derivative of testosterone while ENGA-L06E of estrone. The chemical structures of all studied compounds are shown in Figure 11. human cancers [40]. Nevertheless, the utilization of dual-targeted inhibitors for PI3K/AKT/mTOR and RAS/ERK signaling pathways is a potentially effective therapeutic approach against ovarian cancer [41].
On the other hand, steroid hormones promote epithelial ovarian cancer growth, and their receptor expression is associated with disease outcome. Estrogen (ER-α/-β), and progesterone receptors PR(-A/-B) are frequently expressed in ovarian cancer with an important variability relating to histological subtype, grade, and stage. Hormone therapy is frequently used, but the significance of activity by specific agents or tumor characteristics is unknown. Although the activity of endocrine therapy in advanced ovarian cancer is considerable, limited clinical trials have shown modest response rates to endocrine therapy. However, a small subset of patients that respond very well was identified with few side effects [42][43][44]. The steroid components of ENGA-L06E and ENGA-L08E are testosterone and estrone derivatives, respectively. It has been reported that the steroidal moiety of homo-aza-steroid alkylators is binding to estrogen receptors (ER-α), in breast cancer cells with positive ER expression, and consequently was shown to be more sensitive to their cytostatic and cytotoxic effects [25,45]. However, both ENGA-L06E and ENGA-L08E compounds were shown significant anticancer activity against all human ovarian carcinoma cell lines tested and regardless of the steroid receptor expression.

Synthesis of Chemical Compounds
All modified hybrid steroidal esters and comparative alkylators were synthesized with methods that have been previously described [23]. More specifically, we have tested the two previously synthesized hybrid homo-azasteroid alkylators (e.g., ENGA-L06E and ENGA-L08E) along with the proper reference alkylator aniline mustards (e.g., POPAM and L-PAM). The lactam-steroidal carrier of ENGA-L06E is a derivative of testosterone while ENGA-L06E of estrone. The chemical structures of all studied compounds are shown in Figure 11.

Cell Lines
Five human ovarian cancer cell lines (UWB1.289, UWB1.289+BRCA1, SK-OV-3, OVCAR-3, and OVCAR-5) were treated with the newly synthesized compounds at concentrations of 1-100 µM in order to test their cytostatic and cytotoxic activity. All cell lines were obtained from the American Type Culture Collection (ATCC), USA except OVCAR-5, which was kindly donated from BRFFA, Academy of Athens, Greece, and were grown in different culture media (according to the manufacturer's recommendations), 37 • C and 5% CO 2 . Finally, all specific molecular characteristics and mutated genes' profiles, in all five human ovarian cancer cell lines, concerning chemotherapeutic drug tolerance, are shown in Table A8. The cell lines have been tested with standards procedures for microbial contamination, including mycoplasma.

Determination of Cell Viability by the MTT Assay
Cells were plated in 96-well plate at a density of 1 × 10 4 cells/mL per well and maintained for 72 h at 37 • C in a 5% CO 2 incubator and grown as monolayers. After 24 h, cells were treated with 0.1-100 µmol/L of the tested compounds for 48 h. The viability of cultured cells was estimated by utilizing the (3-(4,5-imethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) metabolic assay, as described previously [23,24]. Briefly, MTT (Sigma, St Louis, Missouri, USA) was dissolved in PBS in a concentration of 5 mg/mL, filter sterilized, and stored at 4 • C. Then, 50 µL of stock solution was added to each culture and incubated for 3 h at 37 • C. Formazan crystals were solubilized by DMSO (100 µL). The absorbance of the converted dye was measured at a wavelength of 540 nm on ELISA reader (VersaMax Microplate Reader Orleans, USA). The mean concentrations of each drug that generated 50% or total (100%) growth inhibition (GI 50 and TGI, respectively), as well as the drug concentrations that produced cytotoxicity against 50% of the cultured cells [half maximal inhibitory concentration (IC 50 )], were calculated using the linear regression method [46]. Using seven absorbance measurements

Determination of ERK 1/2 and AKT 1/2/3 Phosphorylation by ELISA
Phosphorylation of ERK 1/2 (Thr202/Tyr204, Thr185/Tyr187) and AKT 1/2/3 (Ser473) was determined by the Activation InstantOne™ ELISA assay kit (Invitrogen, eBioscience, (San Diego, CA, USA) at 450 nm and according to manufacturer's recommendations. All cell lines were cultured in 12-well plates, seeding 10 5 cells per well. After 24 h, cells were treated with ENGA-L06E and ENGA-L08E, at concentrations of 10, 20, and 60 µM, and time exposures of 0, 15, 30, and 60 min. Cells were washed in PBS, extracted in RIPA buffer, and the protein content was quantified by the Bradford assay. Then, 50 µg/mL of protein concentration was added in a 96-well plate, in triplicates, followed by the addition of 50 µL/well of mixed antibodies AKT/ERK 1,2 and pAKT/ERK 1,2 for 1 h. Samples were washed three times, and then 100 µL/well of detection reagent was added and incubated for 30 min. Finally, the reaction was terminated by the addition of 100 µL/well of STOP solution, after which the absorption of all samples was measured at 450 nm.

Determination of ERK 1/2 and AKT 1/2/3 Phosphorylation by Western Immunoblotting
In order to confirm that ENGA-L06E and ENGA-L08E inhibit the phosphorylation of ERK1/2 and AKT 1/2/3, western immunoblotting was utilized [46]. Cells were treated with the tested compounds for 0, 0.25, 0.5, 1, 3, 6, 12, and 24 h, and samples were prepared in the presence of ice-cold RIPA buffer and protease inhibitor cocktail mixture. The proteinaceous extract was vortexed, centrifuged at 14,000× g, for 10 min, at 4 • C and the total protein content was quantitated by the BCA assay. Then, an equal quantity (25 µg) from each protein sample was loaded into SDS-PAGE electrophoresis followed by transfer to a nitrocellulose membrane with further incubation at 5% skim milk in Tween 20-Tris buffered saline (TBST). After blocking, the membrane was incubated overnight with primary antibodies [ERK1/2, pERK1/2, pAKT, AKT, and Glyceraldehyde 3-phosphate dehydrogenase (GAPDH] at specified concentrations. Finally, the membrane was washed and then incubated with the HRP-conjugated rabbit anti-mouse secondary antibody. Signals were detected using an automated image acquisition system, while all acquired images were further normalized to GAPDH and semi-quantified by the ImageJ software (NIH, Bethesda, Washington D.C., USA).

Determination of In Vivo Toxicity and Antitumor Efficacy
CB17 severe combined immunodeficiency (SCID) male and female mice were obtained from the National Center for Scientific Research (NCSR) Demokritos, Institute of Biology, Athens, Greece. They were fed water and an irradiated standard rodent diet ad libitum and housed under pathogen-free conditions. All animals were sacrificed by cervical disruption when shown signs of cachexia, suffering, or the tumor volume reached 2000 mm 3 . For intraperitoneal (i.p.) treatment, stock solutions of the four tested compounds were prepared immediately before use. They were suspended in corn oil, in the desired concentration, following initial dissolution in 10% dimethylsulfoxide (DMSO) without a record of any observable toxic effects. Briefly, acute toxicity induced by the tested compounds was determined after a single i.p. injection (into groups of ten CB17 SCID mice, which were then observed for 30 days) and as described previously [11,47]. The LD 10 and LD 50 (lethal dose for 10% and 50% of animals, respectively) were determined after graphical estimation (30-day curves) where the percentage of deaths due to the toxicity of each dose was shown in the ordinate while the administered doses were indicated on the abscissa. Two animal models were established based on either SK-OV-3 or OVCAR-5 being implanted subcutaneously (s.c.) into CB17 SCID mice. Overall, at Day 0 (D0), 4 × 10 6 cells were implanted s.c., and mice were randomized into either test (n = 5) or control (n = 7) groups while their tumor volumes were in the 250-300 mm 3 range. After randomization, all animals were treated i.p. with either the tested compounds (at a single LD 10 dose; test group) or saline buffer (at a single 0.2 mL dose; control group). The first day that each animal model was treated with each of the tested compounds was determined as Day 1 (d-1). Finally, all stock solutions for each of the tested compounds were prepared immediately before use. The subcutaneously tumors were measured, and the animals were weighed twice a week starting on the first day of treatment. Tumors were measured using an analog caliper. Tumor volume (mm 3 ) was calculated according to the formula D × d 2 /2, where D and d refer to the larger and smaller perpendicular dimensions collected at each measurement, respectively. The diameters were determined from two orthogonal measurements on each tumor. The mean tumor volume was determined by average, all tumor volumes in each group [48]. On the other hand, for the evaluation of antitumor efficacy, the following parameters were considered:

Elucidation of Underlined Molecular Mechanism(s) by in Silico Approaches
Crystal structures of AKT1, AKT2, ERK1, and ERK2 were available in the RCSB Protein Data Bank of Brookhaven National Laboratory from which we have chosen the ones with the highest resolution (Å) (the higher the resolution, the lower the Å value). The chosen PDB carries a co-crystalized ligand with a carbon chain quite large to resemble as much as possible with our long chain steroid hybrid molecules in order to fit properly at the proteins' binding site cavity. These structures were superimposed, showing no conformational variability for the backbone and for most sidechains. Waters were deleted with Maestro, the graphical interface of Schrödinger software (http://www.schrodinger.com), prior to docking. Molecular docking studies were carried out for the best-fitted compounds to the model, while the final selection criteria were compounds docking scores and the presence of crucial interactions for binding to the studied proteins [50]. Finally, the resulting poses were examined manually, and the most promising ones were used again in docking with Induced Fit Docking protocol 2013-2 [51,52], Glide, version 5.9, Prime, version 3.2 (Schrödinger, LLC, New York, NY) [53,54]. To rationally design the steroidal conjugates to be able to bind to the binding sites of ERK1/2 and AKT1/2, we used for the induced fit docking (IFD) process the different available crystal structures shown in Table A9. All of our newly synthesized compounds are sketched and converted into three-dimensional MOL2 files using Schrödinger Release 2015-2, Maestro Version 10.5, and minimized using Schrödinger Release 2015-2, LigPrep 3.4 [54], and the OPLS3 [20] force field. In all cases, the protonation state for the compounds was adjusted at physiological pH using LigPrep (Schrödinger, http://www.schrodinger.com).

Statistical Analysis
All experiments were repeated in triplicate. Data were presented as mean ± standard deviation (SD). Statistical differences between experimental groups were compared using the Student's t-test (two-tailed paired). Statistical significance was considered at p < 0.05. Microsoft Excell (Microsoft Hellas, Athens, Greece) was used to analyze the data.

Conclusions
Two new homo-aza-steroidal alkylating ester compounds (e.g., ENGA-L06E and ENGA-L08E) were tested in an in vitro model of human ovarian carcinoma (consisting of five cell lines namely UWB1.289, UWB1.289+BRCA1, SK-OV-3, OVCAR-5, and OVCAR-3) as well as an in vivo model consisting of SK-OV-3 and OVCAR-5 tumor xenografts. Both compounds were found to exert very high anticancer activity by inhibiting both the PI3K-AKT and KRAS-ERK signaling pathways. Detailed in silico computational studies were also in accordance and confirmed results of the in vitro studies. As has been previously reported, both of these compounds are capable of inhibiting PARP activity, and consequently, overcome resistance of ovarian cancer cells to alkylating agents, including cisplatin. Moreover, mutations on TP53, KRAS, MEK1 and PIK3CA, microsatellite instability (MSI) status, and steroid receptor expression did not affect the anticancer effect of both compounds. Thus, lactam steroidal alkylators (e.g., ENGA-L06E and ENGA-L08E) provide a novel combination of pharmacological properties exerting significant antitumor activity against human ovarian carcinoma independently of the RAS-ERK and PI3K-AKT mutation, MMR and steroid receptor status, as well as, existing tumor cell resistance to alkylating agents. Conclusively, these data support evidence for further development of lactam steroid alkylators in clinical studies on the treatment of epithelial ovarian carcinoma.

Conflicts of Interest:
The authors declare no conflict of interest.